Tribler: LogParser writes to database BT_HAVE and BT_REQUEST messages

diff --git a/log-parser/tribler/LogParser.py b/log-parser/tribler/LogParser.py
index 51527f7..dc45555 100644 (file)
--- a/log-parser/tribler/LogParser.py
+++ b/log-parser/tribler/LogParser.py
@@ -13,7 +13,7 @@ import getopt
 import re
 
 # the names used by Tribler for the BitTorrent messages 
-msg_types = {"BT_REQUEST": "new_request"} 
+msg_types = {"BT_REQUEST": "new_request", "BT_CHOKE": "_rechoke", "BT_HAVE": "got_have"} 
 msg_db_code = {"BT_CHOKE" : 0, "BT_HAVE" : 4, "BT_REQUEST" : 6}
 
 DEBUG = True
@@ -31,7 +31,21 @@ def usage():
     print "\t--help"
     print "\t-h\t\t\tprint this help screen"
 
-""" line format: 20-10-2009 12:56:39   Downloader: new_request 52 98304 16384 to 141.85.37.41 14398
+""" Get date and timestamp and transform it into datetime format.
+    Format: dd-mm-yyyy hh:mm:ss
+"""
+def tribler_parse_timestamp(date, time):
+    
+    date_array = date.split("-");
+    time_array = time.split(":");
+    if len(date_array) != 3 or len(time_array) != 3:
+        return None
+    
+    timestamp = datetime.datetime(int(date_array[2]), int(date_array[1]), int(date_array[0]), #year, month, day
+                                    int(time_array[0]), int(time_array[1]), int(time_array[2])) #hour, min, sec
+    return timestamp
+
+""" line format example: 20-10-2009 12:56:39   Downloader: new_request 52 98304 16384 to 141.85.37.41 14398
     BitTorrent protocol message: request: <len=0013><id=6><index><begin><length>
     Returns a list (timestamp, peer_ip, peer_port, message_type, index, begin, length, listen_port)
 """
@@ -46,16 +60,11 @@ def tribler_parse_request_msg(line):
         print "Error: invalid line format for Downloader."
         return None
 
-    # get date and timestamp and transform it in datetime format
-    
-    date_array = line_parts[0].split("-");
-    time_array = line_parts[1].split(":");
-    if len(date_array) != 3 or len(time_array) != 3:
+    timestamp = tribler_parse_timestamp(line_parts[0], line_parts[1])
+    if timestamp == None:
         print "Error: invalid line format for Downloader."
         return None
     
-    timestamp = datetime.datetime(int(date_array[2]), int(date_array[1]), int(date_array[0]), #year, month, day
-                                    int(time_array[0]), int(time_array[1]), int(time_array[2])) #hour, min, sec
     index = int(line_parts[4])
     begin = int(line_parts[5])
     length = int(line_parts[6])
@@ -64,13 +73,47 @@ def tribler_parse_request_msg(line):
 
     return (timestamp, peer_ip, peer_port, msg_db_code["BT_REQUEST"], index, begin, length, 0)
 
+
+"""
+    line format example: 20-10-2009 12:56:39 Downloader: got_have 12
+    BitTorrent message format: have <id> <index>
+    Returns a list :...
+"""
 def tribler_parse_have_msg(line):
    
-    pass
+    if line.find(msg_types["BT_HAVE"]) == -1:
+        return None
 
+    # the messages can also have the format: Downloader: got_have <piece_number> is invalid piece
+    if line.find("invalid") != -1:
+        return None # didn't decide yet what to do with it
+    
+    line_parts = re.split(" *", line)
+    timestamp = tribler_parse_timestamp(line_parts[0], line_parts[1])
+    
+    if timestamp == None:
+        print "Error: invalid date & time format for Downloader."
+        return None
+    
+    if len(line_parts) < 5 : 
+        print "Error: invalid line format for Downloader."
+        return None
+
+    index = int(line_parts[4])
+    return (timestamp, None, None, msg_db_code["BT_HAVE"], index, None, None, 0) 
+
+
+
+"""
+    Parses a line that contains choke/unchoke messages.
+    line format: choker: _rechoke
+"""
 def tribler_parse_choke_msg(line):
+    if line.find(msg_types["BT_CHOKE"]) == -1:
+        return None
     pass
 
+
 def tribler_parse_piece_msg(line):
     pass
 
@@ -81,6 +124,7 @@ def tribler_parse_line(line):
            return result
 
     result = tribler_parse_have_msg(line)
+   
     if result != None:
            return result
     
@@ -209,6 +253,7 @@ def main_with_DB():
 
 if __name__ == "__main__":
     sys.exit(main_with_DB())
+    #sys.exit(main_just_parse())